The case families were a cross-sectional series of children with acute leukaemia (n=145) treated at Royal Manchester Children's Hospital, UK, and their parents (n=288; two cases were monozygotic (MZ) twins). Blood samples were obtained from case children at diagnosis and in remission and from their parents. Controls were two series of family trios consisting of unaffected children (n=124) and their parents (n=248). One control series consisted of buccal scrape samples from children (n=64) and blood samples from their parents (n=128), recruited as part of the UK Childhood Cancer Study (2000) in the North West (NW) of England. The second control series consisted of blood samples from children (n=60) obtained as part of the Avon Longitudinal Study of Parents and Children (ALSPAC; Jones et al, 2000) and buccal samples from their parents (n=120; Ring et al, 2001). Sample collection and analyses were carried out with the approval of Local Research Ethics Committees.

CEB1 alleles were amplified by PCR using the 3′ CEB1 primer P14 (5′-ggatcctctcctgtgcctttcct-3′) described by Buard et al (1998) and a 5′ primer, MAR1 (5′-gaattttcagtgagagtcggcc-3′). MAR1 was designed for this study to avoid single nucleotide polymorphisms (SNPs) flanking the CEB1 minisatellite, using the CEB1 sequence (accession no. ) and SNP information kindly provided by Dr Jerome Buard (personal communication). Polymerase chain reactions were performed on an MJ Research DNA Engine (Waltham, MA, USA) in 50 μl reaction mixtures consisting of 1 μl of gDNA (50 ng) and 0.5 μl (2.5 units) of platinum Taq DNA polymerase high fidelity (Invitrogen, Paisley, UK) added to 48.5 μl of Mg²⁺-titrated PCR master-mix. The PCR master-mix (19.4 ml) was prepared with the following final reagent concentrations: 0.5 μM each high-performance liquid chromatography-purified CEB1 primer (Eurogentec, Southampton, UK), 0.7 mM MgSO₄ (50 mM MgSO₄ with platinum Taq DNA polymerase), 0.2 mM dNTPs (PCR nucleotide mix, Amersham Biosciences) 5% (v/v) DMSO (Sigma, , Dorset, UK), 0.2 mg ml⁻¹ bovine serum albumin (ultra pure, non-acetylated; Ambion, Warrington, UK), and 1 × high-fidelity PCR buffer (50 × high-fidelity PCR buffer with platinum Taq DNA polymerase HF). Amplification conditions were 94°C for 1 min 30 s, 30 cycles at 94°C for 15 s, 60°C for 30 s, 70°C for 10 min, and a final extension step of 70°C for 15 min.

Polymerase chain reaction products containing CEB1 alleles amplified from the gDNA of family trios (father, mother, and child) were loaded onto a 40-cm long 0.7% (w/v) agarose gel (SeaKem LE Agarose, Cambrex BioScience, East Rutherford, NJ, USA). Gels were run at 2 V cm⁻¹ in 1 × TBE buffer (Crystal Buffers, Severn Biotech, Kidderminster, UK) for approximately 23 h, until the 400 bp marker from the DNA ladder (1 kb plus DNA Ladder, Invitrogen) was about to run off the gel. Few alleles <400 bp were observed, but when present were sized by a shorter gel run. Multiple ladder lanes were included across each gel to enable the image analysis software to calculate the gel retardation factor (R_f value) for any region of the gel, allowing more accurate allele sizing. Gels were stained with SYBR gold nucleic acid gel stain (Molecular Probes) in 1 × TBE buffer (pH 8.0), the bands being visualised by UV transillumination on a Gel Doc 2000 instrument (Bio-Rad, Hemel Hempstead, UK). The captured gel image in TIFF file format was set up with an image resolution of 38.4 pixels cm⁻¹.

Allele typing and mutation detection were performed on the gel image using Phoretix™ 1D Advance software (Nonlinear Dynamics, Newcastle upon Tyne, UK). Offspring bands (alleles) were deemed to be germline mutations if no corresponding alleles were detected in either parent. Band shifts were measured in bps so that allele sizes could be adjusted for variations in R_f. This was carried out by placing a simulated band on the gel image at a position 1.5 mm (corresponding to ∼8 pixels) from the band of interest, in the direction of the progenitor band in the parent. The progenitor band was taken to be the parental allele closest in size to the mutant band (Dubrova et al, 1996). If the size of the putative mutant allele was equal to or larger (for an expansion), or equal to or smaller (for a deletion), than the simulated band (as measured in bp), it was classified as a mutation. All mutations detected were verified by repeat analysis, in which gDNA was re-amplified and run on a second gel.

Biological parentage was verified in case and control families from NW England with suspected germline mutations using three short tandem repeat (STR) markers (VWA, TH01, and FGA) from the FBI combined DNA index system (CODIS) and the minisatellite, D1S80, as described in the National Institute of Standards and Technology STR DNA internet database (http://www.cstl.nist.gov/biotech/strbase/; Ruitberg et al, 2001) and using HLA-DPB1 typing data already available for families (Taylor et al, 2002). Primer sequences for STRs were: VWA 5′-ccctagtggatgataagaataatcagtatg-3′ and 5′-ggacagatgataaatacataggatggatgg-3′; TH01 5′-attcaaagggtatctgggctctgg-3′, 5′-gtgggctgaaaagctcccgattat-3′; and FGA 5′-attatccaaaagtcaaatgccccatagg-3′, 5′-atcgaaaatatggttattgaagtagctg-3′. Primer sequences for D1S80 were: 5′-gtcttgttggagatgcacgtgccccttgc-3′, 5′-gaaactggcctccaaacactgcccgccg-3′ (Kasai et al, 1990). High-performance liquid chromatography-purified primers were obtained from Eurogentec or Applied Biosystems (Warrington, UK), the forward primer for each set being labelled at the 5′-end with VIC (VWA), NED (TH01), or 6-FAM (FGA and D1S80). All four loci were co-amplified and analysed on an ABI-310 genetic analyser (Applied Biosystems). Reagent concentrations in 20 μl reaction volumes were 1 × ReddyMix™ PCR master mix with 2.5 mM MgCl₂ (ABgene, Epsom, UK), 0.25 μM each primer and 2.5 ng μl⁻¹ DNA. Polymerase chain reaction amplification conditions were 95°C for 2 min, 30 cycles at 95°C for 30 s, 63°C for 30 s, 70°C for 2 min, and a final extension step of 70°C for 5 min. Parentage was verified in mutation carriers if alleles in at least two of the four tandem-repeat loci were present in parents; DP types were used as an independent check. Parentage of all ALSPAC control family trios was verified using four STR loci (D5S818, D7S820, D13S317, D16S539; http://www.cstl.nist.gov/biotech/strbase/) by multiplex PCR using GammaSTR kits (Promega, Southampton, UK) as described by Ong et al (2004).

Case–control CEB1 germline mutation rates were compared by calculating cross-product odds ratios (ORs) and 95% confidence intervals (CI) using the RERI program from the linkage utility package, LINKUTIL, by the Sheehe method (http://linkage.rockefeller.edu/soft/linkutil/). The 2 × 2 programme in LINKUTIL was used to determine P values for case–control differences using Fisher's exact test. Mean case–control parental ages and germline progenitor allele sizes were compared using unpaired t-tests by Minitab release 13.30. Case–control parental allele size distributions were compared using the Kolmogorov–Smirnov Z two-independent samples test. Statistical power was calculated using nQuery Advisor v2.0 (Statistical Solutions, Cork, Ireland).

A total of 145 leukaemia cases from 144 families were originally included in the study. Two cases with B-cell precursor ALL (BCP ALL) were MZ twins and were treated as one case in the analysis. Control families consisted of 124 unaffected children and their parents. Nine case families were excluded from the study; five because of failure to detect CEB1 alleles; four because of non-parentage, leaving 135 informative case families. Table 1 shows the number of informative case children with each leukaemia subtype. B-cell precursor ALL is the predominant subtype (70%), the majority (68%) of case samples were collected at diagnosis. The mean age of cases at diagnosis of leukaemia was 5 years 5 months (range: 0.2–15.0 years). The mean paternal age at the birth of case children was 30.4±(s.d.) 5.4 years; for the control series were the following, NW: 31.0±6.3 years; ALSPAC: 32±5 years. Mean maternal ages were: case children 28.5±5.7 years; NW controls: 29.1±4.6 years; ALSPAC controls: 30±4.1 years. There were no significant differences between case and control parental ages (P>0.05)

CEB1 genotyping was carried out by amplifying gDNA in PCR to levels visible in SYBR gold-stained gels. A very precise Mg²⁺ concentration (to within 0.1 mM) was found to be required for successful amplification and allele discrimination (Figure 1). Using this method, we obtained an allele amplification efficiency of 90%, with no amplification artifacts and consistent results in repeat sample analysis.

Germline mutations were detected in 11 of 135 case and 20 of 124 control children. Representative examples of mutations (one expansion and one deletion) in two case families are shown in Figure 2. All case mutations were paternal in origin; two of 21 control mutations were maternal. One control but no cases had a double, paternal and maternal, mutation. Table 2 shows that there is no significant difference in parental mutation rates in case (0.041) and control (0.086) families (OR, 95% CI: 0.47, 0.23–0.98; P=0.061). Paternal mutation rates in case (0.083) and control families (0.156) were also not significantly different (OR, 95% CI: 0.50, 0.23–1.08; P=0.11). Analysis of mutation rates in relation to leukaemia subtype (Table 3) shows no significant case–control differences, though the small number of pro B and T ALL and acute myeloid leukaemia (AML) suggests that these results should be treated with caution. The subtype with the highest mutation rate was T ALL (0.111), but this was not significantly different from controls (OR, 95% CI: 0.8, 0.22–3.00; P=0.93). The mean age of case fathers with a germline mutation (29.9 years) was not significantly different from NW control fathers with a mutation (33.9 years; P=0.27). The mean age of case fathers whose children had a germline mutation (29.9 years) was not significantly different from fathers of case children who had no mutation (30.4 years; P=0.81). The mean germline mutation rate in diagnostic samples (0.039) was not significantly different from remission samples (0.046; P>0.05). The CEB1 allele size distributions (Figure 3) in case parents (mean allele size: 2865 bp; 72 repeats) was not significantly different from control parents (mean allele size: 2360 bp; 59 repeats; P>0.05).

In cases, there were seven (64%) mutant alleles with an increased and four with a decreased number of repeats, whereas in the controls, 15 (71%) mutants increased and six decreased in size. The size of the mutational events was similar in the cases and controls. The mean mutant allele size in cases (2.71 kb) was not significantly different from controls (2.67 kb; P<0.90). There was no difference between mean progenitor allele size in cases (2.74 kb) and controls (2.54 kb; P=0.56). Most mutations (75%) involved the gain or loss of 1–5 repeat units (Figure 4), but there was no significant difference in the size change in cases (0.13 kb) and controls (0.25 kb; P=0.10). The largest control mutation was an expansion of 32 repeat units (progenitor allele size: 2.09 kb); the largest case mutation was a deletion of nine repeats (progenitor allele size: 2.55 kb) in a child with BCP ALL.